[mpiwg-sessions] Looks like we do have a topic to discuss soon

[Holmes, Daniel John]

Hi Howard,

The MPI_SESSION_INIT procedure is listed as “l” for local and (2) meaning not related to an MPI operation in the huge semantics table in Appendix A (MPI-4.1, p 881, line 29).

This is correct but quite well hidden. Perhaps we could add “This procedure is local.” in normative text just after the binding?

We do elucidate the point, using an example (Example 11.9) where two MPI objects (communicators) are created using the same session in one MPI process but in different sessions in another MPI process. This example cannot be correct unless the creation/instantiation of a session is NOT collective and does NOT synchronise.

We could also include one or more pictures, like the ones we’ve used in the WG when talking about bubbles. The text art in Example 11.9 is not the most pleasing visual expression of the concept.

Best wishes,
Dan.

From: mpiwg-sessions <mpiwg-sessions-bounces at lists.mpi-forum.org> On Behalf Of Pritchard Jr., Howard via mpiwg-sessions
Sent: Friday, July 12, 2024 4:32 PM
To: MPI Sessions working group <mpiwg-sessions at lists.mpi-forum.org>
Cc: Pritchard Jr., Howard <howardp at lanl.gov>
Subject: [mpiwg-sessions] Looks like we do have a topic to discuss soon

HI Everyone, Well it looks like we have a topic we do need to discuss soon, possibly for an errata item for 4. 2. There seems to be more ambiguity about what is implied by the wording in section 11. 3 concerning MPI_SESSION_INIT – Session instantiation
ZjQcmQRYFpfptBannerStart
This Message Is From an External Sender
This message came from outside your organization.
ZjQcmQRYFpfptBannerEnd
HI Everyone,

Well it looks like we have a topic we do need to discuss soon, possibly for an errata item for 4.2.

There seems to be more ambiguity  about what is implied by the wording in section 11.3 concerning
MPI_SESSION_INIT – Session instantiation is intended to be a lightweight operation than many of us may have realized.

We don’t explicitly state whether this call is in any sense synchronizing with other processes.

We’ve talked about this so long, for instance in terms of asymmetry with MPI_SESSION_FINALIZE, that we didn’t consider
that an implementor may construe that MPI_SESSION_INIT has the same implicit synchronization behavior as MPI_INIT.

We may want to add something in an errata to clarify what lightweight means.

Again I will not be able to call in on Monday but perhaps if others want to discuss?

The meeting will be automatically recorded BTW.

Thanks,

Howar


-------

[signature_61897647]
Howard Pritchard
Research Scientist
HPC-ENV

Los Alamos National Laboratory
howardp at lanl.gov<mailto:howardp at lanl.gov>

[signature_4180174387]<https://urldefense.us/v3/__https:/www.instagram.com/losalamosnatlab/__;!!G_uCfscf7eWS!dbZccRYCcJZ8TbF_EhT6HopcdaJyb4u97DQfUDmGDEhNCAAU2hpfMv6bIBhU3PWR2PzW6MZVJ2di947WgOPbI6wjOnz_WCc$>[signature_937381201]<https://urldefense.us/v3/__https:/twitter.com/LosAlamosNatLab__;!!G_uCfscf7eWS!dbZccRYCcJZ8TbF_EhT6HopcdaJyb4u97DQfUDmGDEhNCAAU2hpfMv6bIBhU3PWR2PzW6MZVJ2di947WgOPbI6wjXwr6J60$>[signature_35314574]<https://urldefense.us/v3/__https:/www.linkedin.com/company/los-alamos-national-laboratory/__;!!G_uCfscf7eWS!dbZccRYCcJZ8TbF_EhT6HopcdaJyb4u97DQfUDmGDEhNCAAU2hpfMv6bIBhU3PWR2PzW6MZVJ2di947WgOPbI6wjRk6GtnI$>[signature_2948438747]<https://urldefense.us/v3/__https:/www.facebook.com/LosAlamosNationalLab/__;!!G_uCfscf7eWS!dbZccRYCcJZ8TbF_EhT6HopcdaJyb4u97DQfUDmGDEhNCAAU2hpfMv6bIBhU3PWR2PzW6MZVJ2di947WgOPbI6wjG2gdL6U$>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 4350 bytes
Desc: image001.png
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 1981 bytes
Desc: image002.png
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image003.png
Type: image/png
Size: 1517 bytes
Desc: image003.png
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0007.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image004.png
Type: image/png
Size: 1334 bytes
Desc: image004.png
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0008.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image005.png
Type: image/png
Size: 999 bytes
Desc: image005.png
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20240712/18948e2b/attachment-0009.png>